Getting specific with biomarkers
Freelance science writer Jim Kling reports:
At a Sunday session titled "Next Generation in Diagnostics," the talk turned to biomarkers. The panelists agreed that no one has a good handle on how well a biomarker must reflect a disease state in order to be clinically useful.
Anthony Shuber, CTO of Predictive Biosciences lamented that association studies are appearing constantly in the literature, but there is rarely any understanding of the underlying biology -- that is, an understanding of the biological consequences of an up- or down-regulated gene, or
the concentration level of a protein. The issue is specificity. Is the gene upregulation or increase in protein level specifically associated with a disease state, or does it also appear in some normal states, or perhaps related or even unrelated disease states? To determine that, companies will have to carry out large, expensive clinical trials to demonstrate a biomarker's specificity.
An understanding of the underlying biology could circumvent that. "If you understand the biology, you can go in front of an FDA panel and say, 'this is the gene, this is what it does. Therefore it's a valid biomarker. There should be no need to do a large specificity study," Shuber said.
Another panelist suggested that what we need are molecular biologists who understand physiology, who can make connections between associations and known biological mechanisms, as well as biostatisticians who know how to distinguish a valid biomarker from the biological 'noise' that permeates any cell system.
